A Design of Continuous Learning System Based on Knowledge Augmentation
Hyunjoong Kang, Soon Hyun Kwon, Eun Joo Kim, HyunJae Kim, Ho Sung Lee, Kwihoon Kim, Nae-Soo Kim · 2017
To create an algorithm with Machine Learning, users should understand all the knowledge such as learning rate, activation, dimension reduction, hyper parameter, neural network, etc. Therefore, in order to construct a machine learning procedure, expert knowledge is required. So, it is difficult for general users to use it. Also, experts are also hard to regenerate well-defined model if it is described only in the paper. In this paper, we propose a knowledge based Continuous Learning System (CLS) which persistently collect and infer new knowledge from information for the existing learning setup and results instantiated based on a hierarchically designed ontology model.